Objective Randomised controlled trials are the gold standard for evaluating interventions but especially in occupational health not always feasible. Therefore, non-randomised studies (NRS) are increasingly used as evidence for effectiveness of interventions also in Cochrane reviews. When and how NRS are included has not been evaluated to date. Our aim was to conduct an overview of practice to show what kinds of questions are addressed, what kind of methods are used and what reasons the review authors cite for the inclusion of NRS within the Cochrane Collaboration.
Methods We searched the Cochrane Database of Systematic Reviews (CDSR). We included all reviews that aimed to include NRS. We conducted study selection and data collection in duplicate and analysed the results with ATLAS.ti and Excel. We analysed how questions were addressed and reasons for inclusion were distributed over review groups, study participants and interventions.
Results We included 202 reviews. The earliest reviews were from the year 2000. The number of Cochrane reviews with NRS has consistently increased over the years. Most of the reviews (52%) did not cite a reason. Where cited the most common reason for inclusion of NRS was non-feasibility of RCTs for an intervention (30%). It was not always clear why RCTs were not feasible. The highest number of reviews with NRS (61) came from the EPOC group. The reviews mostly addressed health care providers (28%). The most common tools for risk of bias assessment were EPOC group’s criteria (28%) followed by The Cochrane risk of bias tool (15%). The assessment was not described in 3% of the reviews.
Conclusions Reasons for including NRS in systematic reviews vary across Cochrane review groups. Reasons for non-feasibility of RCTs should be better elaborated. Definition of study designs and risk of bias assessment of NRS needs more attention.